A small RNA promotes siderophore production through transcriptional and metabolic remodeling

The production of siderophores from E. coli K-12 was monitored for WT and ΔryhB cells growing in minimal M63 medium with or without addition of 1 μM FeSO₄. In absence of Fe, the WT strain produces a considerable amount of siderophores (Fig. 1A, lane 2), which migrates to the same level as the purified enterobactin control (lane 1). In contrast, we observed a dramatic decrease in siderophore production in the ΔryhB mutant even in the absence of Fe (Fig. 1A, lane 4). This is unexpected because, to our knowledge, almost every gene involved in siderophore biosynthesis and export is known to be regulated at the promoter level by Fur (30). These data indicate that even in low Fe conditions, when Fur is inactive, RyhB has a critical role in biosynthesis and/or secretion of siderophores.

Because the expression of siderophore genes is strongly linked to the Fur regulon, we examined the effect of a Δfur mutation in a ΔryhB mutant. Our results indicate that inactivation of fur in a ΔryhB background restores the production of siderophores to a level similar to that of the WT (Fig. 1A, compare lanes 2 and 8). Nevertheless, the double Δfur ΔryhB mutant produces significantly fewer siderophores than the Δfur mutant (Fig. 1A, compare lanes 6 and 8). This suggests that even in the absence of fur, when siderophore biosynthesis genes are fully derepressed, RyhB expression still enhances the production of enterobactin. These results demonstrate a specific role for RyhB in the production of siderophores. Notably in the conditions we used, the growth curves for all cells were comparable with or without Fe supplementation as determined on a bioscreen (Fig. S2).

To investigate the role of RyhB in enterobactin production, we quantified the levels of enterobactin siderophores produced from cells growing in increasing amounts of Fe. To do this, we measured siderophores directly from culture supernatants using liquid chromatography coupled with mass spectrometry (LC–MS). The amounts of siderophores produced during growth without Fe (Fig. 1B) are consistent with the data observed by TLC (Fig. 1A lanes 2, 4, 6, and 8). When Fe was added at 1 μM in the culture, the WT strain shows reduced siderophore production (Fig. 1B), which is consistent with the conventional idea that more Fe enables Fur to shut down ent genes. As expected, when RyhB is expressed from a plasmid vector in ΔryhB cells, the production of siderophores becomes similar to that of WT cells (Fig. 1C, pRS-c25).

The reduction of siderophores observed in a ΔryhB mutant raises the possibility that, without RyhB, the free intracellular Fe becomes sufficiently elevated to activate the Fur repressor. Thus, we monitored the intracellular concentration of free Fe by using electron-paramagnetic resonance (EPR) on various cells (31). Our results demonstrate that, when ΔryhB cells grow in M63 medium without Fe, they have about 3-fold less free intracellular Fe than WT cells (Fig. 2, compare WT and ΔryhB, without Fe). Therefore, we cannot conclude that siderophore production decreases in the ΔryhB mutant because of increased free Fe and Fur activation. However, this confirms that RyhB increases free intracellular Fe levels in the WT background growing under Fe starvation (Fig. 2, compare columns 1 and 3), demonstrating the Fe-sparing activity of RyhB.

In addition, although the Δfur cells (column 5) have about 4-fold more free intracellular Fe than WT cells (column 1), the double Δfur ΔryhB cells (column 7) have as little free Fe as the ΔryhB cells (column 3) when grown without Fe. This demonstrates the strong effect of RyhB on the levels of free intracellular Fe. However, when 1 μM Fe was added to the culture, both the ΔryhB and the Δfur ΔryhB cells (columns 4 and 8) demonstrated a dramatic increase in free intracellular Fe. In contrast, the WT cells do not show any significant variation whether Fe is present or not (columns 1 and 2), demonstrating the robust mechanism for maintaining intracellular Fe homeostasis when both Fur and RyhB are functional.

As shown in Fig. 2, the free intracellular Fe is dramatically decreased in a ΔryhB mutant as compared with WT under low Fe growth conditions. This low Fe availability leads one to expect that, in the ΔryhB mutant, the Fur repressor must be inactive and Fur-regulated genes are fully derepressed. This was tested by monitoring the mRNA level of several genes involved in siderophore synthesis (entC, entB, and entA), secretion (entS and tolC), uptake (fepA), and transcriptional regulation (fur). In fact, the results in Fig. 3A clearly indicate that not only entC, entB, and entA genes are not derepressed but they are significantly reduced in the ΔryhB background as compared with WT at OD₆₀₀ of 0.9 and 1.2 (ratio ΔryhB/WT <1). In contrast, both entS and tolC mRNAs remained equally expressed whether RyhB is present or not (ratio ΔryhB/WT ∼1). Although entC, entB, entA, fepA, entS, and fur are Fur-regulated genes, tolC expression is independent from Fur. Because entS and fur are not affected in these conditions, we cannot conclude that all Fur-regulated genes are repressed in ΔryhB cells. Thus, even if intracellular Fe is low enough to inactivate Fur repression, we nevertheless observe significant repression of many Fur-regulated genes in ΔryhB cells.

The previous results in Figs. 2 and 3A demonstrate that even though intracellular free Fe is low in a ΔryhB mutant, it is not sufficient to induce Fur-regulated genes. This suggests that Fur is still active under low Fe in a ΔryhB mutant. Indeed, RyhB was previously shown to partially repress fur translation (29). Thus, when growing under Fe starvation we expect Fur protein level to be higher in ΔryhB cells as compared with WT cells. However, as shown in Fig. S3, a Western blot performed from cells expressing RyhB (WT) or not (ΔryhB) indicates that the Fur protein levels are similar in ΔryhB background as compared with WT. This demonstrates that RyhB does not affect fur translation in WT cells growing under Fe starvation. These data suggest that increased Fe, through the Fe-sparing action of RyhB, plays a key role in activating the expression of enterobactin synthesis genes. We tested this hypothesis by monitoring the effect of Fe on the levels of a number of transcripts in ΔryhB Δfur cells, which should not have any Fe-dependent effectors. As shown in Fig. S4, the mRNA levels of genes related to enterobactin, namely entCEBAH, entS, and fepA are significantly higher in ΔryhB Δfur cells in the presence of 1 μM FeSO₄. Thus, when growing under Fe starvation, RyhB-expressing cells will have increased intracellular Fe (Fig. 2, compare columns 1 and 3), which may contribute to the transcriptional activity or transcript stability of genes involved in synthesis of enterobactin.

The entCEBAH polycistron is required for the synthesis of DHB, which is one of the building blocks for enterobactin synthesis (Fig. S1). If entCEBAH expression is reduced in a ΔryhB mutant, then DHB may not be sufficiently synthesized to allow siderophore production. To investigate this, we supplemented DHB in our ΔryhB culture and monitored siderophore synthesis. As shown in Fig. 3B, the addition of 5 μM DHB to WT and ΔryhB cells increases dramatically the production of siderophores specifically in ΔryhB cells. This demonstrates that DHB is limiting in ΔryhB cells, most likely due to reduced expression of siderophore synthesis genes (entCEBAH), as shown in Fig. 3A.

Although an explanation for the low entCEBAH expression in ΔryhB cells remains elusive, another role for RyhB in regulation of siderophore synthesis was observed. In addition to DHB, the synthesis of enterobactin depends on the availability of the amino acid serine as a substrate. Serine is added to DHB through the nonribosomal peptide synthesis (9) in the final steps of the synthesis pathway to form enterobactin (Fig. S1). By using the bioinformatic tool TargetRNA (32), we observed that the gene cysE, encoding serine acetyltransferase, is a potential mRNA target of RyhB (see pairing in Fig. 4A). Serine acetyltransferase converts serine to O-acetyl-l-serine as the first step in the synthesis of cysteine. Because serine acetyltransferase (CysE) activity could limit the availability of serine for enterobactin synthesis, we reasoned that, under low Fe conditions, RyhB might reduce cysE expression to retain sufficient serine for enterobactin assembly. To address this, we used in vitro RNase T1 (cleaves unpaired guanines) and TA (cleaves unpaired adenines) and lead acetate (PbAc, cleaves any unpaired residues) assays for RyhB pairing with cysE mRNA. As demonstrated in Fig. 4B, RyhB clearly pairs at the ribosome-binding site of cysE mRNA. This result suggests that RyhB pairing with cysE mRNA inhibits translation initiation. We tested this hypothesis by using in vitro toeprint assays. As shown in Fig. 4C, the presence of RyhB (lane 11) clearly blocks the binding of the 30S ribosome subunit on the cysE mRNA as compared with cysE without RyhB (lane 7). These results strongly suggest that RyhB specifically binds to cysE mRNA to reduce translation initiation.

We then monitored the cysE translation activity by using a protein fusion with the lacZ reporter gene in vivo. As shown in Fig. 5A, the expression of the cysE′–′lacZ translational fusion is significantly reduced in the WT background as compared with ΔryhB cells. This demonstrates that RyhB efficiently reduces CysE protein level, and most likely serine acetyltransferase activity. To confirm pairing in vivo, we expressed either RyhB or the mutated RyhB6 construct from an arabinose-inducible vector and monitored the effect on WT cysE′–′lacZ and mutated cysE6′–′lacZ fusions (described in Fig. S5). As shown in Fig. 5B, the expression of the RyhB6 affects only the complementary cysE6′–′lacZ construct without affecting the WT cysE′–′lacZ fusion. This demonstrates that RyhB directly represses cysE translation in vivo.

Finally, we monitored by quantitative real-time PCR the cysE mRNA level after a 10-min pulse expression of RyhB. This pulse expression limits the possible indirect effect of expressing a sRNA. As shown in Fig. 5C, the mRNA level of cysE is significantly lower in the presence of the sRNA (pBAD–ryhB) as compared with a strain without RyhB (pNM12) or a control nontarget mRNA such as icd (25).

The previous results indicated that RyhB reduces cysE mRNA and protein levels. This suggests that a high level of CysE is incompatible with siderophore synthesis. To address this, we monitored the siderophore production from a strain carrying the cysE gene and endogenous promoter on a multicopy pBR322-derivative plasmid to overproduce CysE. As shown in Fig. 6A, the level of siderophore production is significantly less (30%) in the strain carrying the multicopy cysE (pFRΔ–cysE) gene as compared with the empty control vector (pFRΔ). We do not expect a full repression of siderophore production because the ryhB gene is present in this background. These data suggest that overexpression of CysE reduces the cell's ability to produce siderophores.

If cysE expression is too high in the ΔryhB cells, then mutating the cysE gene should restore conditions in which siderophores are readily produced. Thus, we monitored siderophore production by a ΔryhB ΔcysE double mutant. As expected, the production of siderophore by a ΔryhB ΔcysE mutant is restored to normal WT levels (Fig. 6B). This shows the importance of keeping the expression of both RyhB and cysE in balance for normal siderophore production. Because both ΔcysE and ΔryhB ΔcysE cells must be supplemented with cysteine to grow, we monitored the effect of the addition of cysteine (125 μM) on the production of siderophores from WT and ΔryhB cells. As shown in Fig. S6, there is no significant effect of the addition of cysteine on the production of the siderophore enterobactin and derivatives.

Enterobactin was extracted and visualized according to a previous report (12). Cells were grown in M63, 0.2% glucose, from a 10-fold or 100-fold dilution of an overnight culture in the same media. Depending on the experiment, 33 μM of DHB or 1 μM of FeSO₄ were added to the media. At an OD₆₀₀ of 0.9, 4 mL of culture was pelleted. The supernatant was acidified with 25 μL of 10N HCl and extracted twice with a total of 4 mL of ethyl acetate. Aqueous phases were combined and dried in 874-μL aliquots in a SpeedVac Concentrator (Savant Instruments, SVC100). Extract residues were resuspended in 40 μL of methanol and 10 μL was spotted onto 250-μM layer-flexible (20 × 20 cm) PE SIL G/UV254 plates (Whatman). For some experiments, 25 μL of enterobactin (EMC Microcollections) was spotted as a control onto the plates. Plates were developed with benzene:glacial acetic acid:water (125:72:3 vol/vol/vol) in a closed chamber. Plates were then removed from the chamber and allowed to dry, then immersed briefly in 0.1% FeCl₃ to visualize Fe-binding compounds.

Cells were grown overnight at 37 °C in M63 glucose medium, diluted 1:10 into 250 mL of freshly prepared M63 0.2% glucose ±1 μM of FeSO₄ and then grown at 37 °C in 1-L baffled flasks with vigorous shaking to an OD₆₀₀ of 0.9. Cells were then harvested and prepared for EPR analysis as described in previous studies (28, 35), with the exception that M63 medium with 0.2% glucose was used in place of LB during the incubation of cells with DTPA and desferrioxamine.

Enzymatic and chemical probing experiments were performed as described earlier (36). Briefly, 50 pmol of cysE mRNA (transcribed from a PCR product—oligos EM1056–EM1153) was labeled using T4 polynucleotide kinase (New England Biolabs). Then, 0.1 μM of 5′-end radiolabeled cysE was incubated 15 min at 37 °C in the absence or in the presence of 1.6 μM RyhB RNA (transcribed from a PCR product—oligos EM88–EM89). Then, RNase T1 (0.05 U) (Ambion), or RNase TA (0.025 U) (Jena Bioscience), or PbAc (10 mM) (Sigma-Aldrich) were added to the reaction and the incubation continued for 2 min. Reactions were stopped by adding 10 μL of loading buffer II (Ambion). Samples were then separated on a 6% polyacrylamide/7 M urea gel.